Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Verify ctx changes — measured AI agent skill | DecimalAI

stevesolun/ctx-verify

NewSkillScoreMaintained

Select proportional validation for changes or reviews in the ctx repository. Use before handing off material code, workflow, migration, documentation, security, packaging, or release changes; before a PR; or when asked whether a change is adequately tested. Do not use for pure questions with no changed artifact.

Testingby stevesolun

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

QA & Test Engineering pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/stevesolun-ctx-verify/SKILL.md

decimalai skills pull stevesolun-ctx-verify --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/stevesolun-ctx-verify/SKILL.md https://app.decimal.ai/s/stevesolun-ctx-verify/SKILL.md

Raw file: /s/stevesolun-ctx-verify/SKILL.md ↗ · machine-readable summary at /s/stevesolun-ctx-verify.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 4h ago

—-43% tokens0% turns55% of whole cases pass with skill (12/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-09	✗→✓	▲ Improved	-45%	0%
case-11	✗→✓	▲ Improved	-35%	0%
case-15	✗→✓	▲ Improved	-37%	0%
case-16	✗→✓	▲ Improved	-33%	0%
case-19	✗→✓	▲ Improved	2%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

stevesolun / ctx

.claude/skills/ctx-verify/SKILL.md

View on GitHub ↗

License MITImported 8/3/2026Version v1

Skill contents

SKILL.md

Verify ctx changes

Inspect the requested behavior and changed surface. Read only the relevant

section of CONTRIBUTING.md before selecting checks.

Run the smallest focused tests or deterministic checks that exercise the

changed behavior first.

Escalate evidence in proportion to risk:

Small, localized change: focused tests, schemas, or diff checks.
Material repository change: scripts/no_mistakes_run.sh fast.
PR-level, contract, workflow, migration, packaging, security, or release

change: run the fast gate, then python scripts/ci_preflight.py --profile pr.

Documentation navigation or linked public-asset change: follow the tracker

contract in CONTRIBUTING.md; do not require tracker changes for ordinary prose-only edits.

Windows-sensitive change: follow the native-CI contract in

CONTRIBUTING.md; local checks do not replace required Windows evidence.

Treat the fast gate as committed-history evidence. When the working tree is

dirty, pair it with focused checks against the actual uncommitted changes.

If a gate cannot find its trusted pytest, Ruff, and mypy environment, report

the environment failure. Do not silently install dependencies or weaken the trust check.

Use an independent validator only when semantic risk remains after

deterministic checks, especially for material security, migration, or release work.

Report each check, its result, and any omitted or unavailable evidence.

Rubric

Mark each applicable criterion pass, fail, or not verified:

Intent: The artifact satisfies the request without unrelated scope.
Behavior: Fresh evidence covers the changed success path and relevant

failure or regression paths.

Static integrity: Applicable lint, formatting, and type checks pass.
Repository contracts: Migration, changed-file, documentation, packaging,

and platform requirements remain satisfied where relevant.

Evidence quality: Results apply to the actual changed surface and are not

stale or inapplicable.

Handoff: Skipped checks and residual risks are stated plainly.

Do not claim completion while an applicable criterion fails. Treat missing required evidence as not verified, not as a pass.

Want to see if stevesolun/ctx-verify would activate on your phrasing? Try it before you install — no LLM call, no signup.

Other measured skills in the registry, with their headline benchmark lift.